Wu 2008b.

Study characteristics
Methods	Study design: Randomized controlled trial Study grouping: Parallel group How were patients recruited: From outpatient clinical of Mental Health Institute between 2004 and 2006. Type of RCT: Double blind
Participants	Baseline characteristics Lifestyle intervention + metformin Sex(N, % female, % male): female 17 (53%), male 15 (47%) Age (range and mean, SD): 24.3 to 27.8, mean 26.1 years Number of participants: 32 Metformin Sex (N, % female, % male): female 16 (50%), male 16 (50%) Age (range and mean, SD): 25.2 to 28.3, mean 26.8 years Number of participants: 32 Lifestyle intervention + placebo Sex (N, % female, % male): female 15 (47%), male 17 (53%) Age (range and mean, SD):24.8 to 28.1, mean 26.4 years Number of participants: 32 Placebo Sex N, % female, % male): female 17 (53%), male 15 (47%) Age (range and mean, SD):24.1 to 27.6, mean 25.8 years Number of participants: 32 Overall Sex (N, % female, % male): female 64 (50%), male 64 (50%) Age (range and mean, SD): 25.5 to 27.1, mean 26.3 years Number of participants: 128 Inclusion criteria: Age 18 to 45 years, first psychotic episode of schizophrenia (DSM‐IV), weight gain over 10% in the year after medication started, weight and antipsychotic treatment history available, stable improvement PANSS, taking only one antipsychotic agent with no major dose change in last 3 months. Exclusion criteria: Liver or renal dysfunction, cardiovascular disease, diabetes, pregnant or lactating, other conditions limiting ability to perform lifestyle modifications, psychiatric diagnosis other than schizophrenia or history of substance abuse. Pretreatment: No statistically significant differences at baseline.
Interventions	Intervention characteristics Lifestyle intervention + Metformin Class of drug: Behavioural intervention + Antidiabetic Dose: 250 mg, increased after 4 days to 750 mg. Behavioural intervention dose unclear Frequency: Metformin: Once a day, increased to 3 times a day. Behavioural intervention: during 4 weeks + daily exercise Duration: 12 weeks Who delivered the intervention: Unclear Type of intervention: Pharmacological + Behavioural Description of the intervention: The lifestyle interventions included psychoeducational, dietary, and exercise programs. Target group was adult patients with schizophrenia. informing intervention design: For the dietary intervention, the American Heart Association (AHA) step 2 diet was prescribed. provider and Mode of delivery: participants maintained 3‐day food records before each follow‐up visit. The dietitian reviewed the food records, compared them with the prescribed AHA diet, and discussed with each patient concerns about adherence to the AHA diet. For the first week of the study, exercise sessions were directed by an exercise physiologist. Participants performed endurance exercise (walking or jogging) on a treadmill 7 times a week for 30 minutes at each session. Exercise was designed to attain 70% of heart rate reserve [0.7(maximum heart rate –resting heart rate)resting heart rate]. Heart rate reserve was estimated from the maximum and resting heart rates observed at baseline maximal aerobic capacity (V˙ O2max) for each individual. After the first week, exercise was home‐based without supervision by investigators. Therapists and patients collaboratively developed individual programs of gradual assignment of exercise. A range of different types of exercise and various levels of intensity was offered, varying from light exercise (increase walking and decrease sedentary activities, particularly time spent watching television or sleeping) to moderate exercise for at least 30 minutes per day. Common moderate‐to‐vigorous physical exercise included walking at a moderate or very strenuous intensity, bicycling, resistance training, skiing, jogging, ball games, and lifestyle activities, such as chopping wood or clear. Adherence: Branching treadmill tests were performed under supervision at each follow‐up visit. Good exercise adherence was defined as an increased V˙ O2max of at least 1.5 mLkg−1.min−1 at 4, 8, and 12 weeks compared with baseline. between 50% and 57% had V˙ O2max results that demonstrated adherence to exercise. The participants’ adherence to metformin treatment for each visit was defined as taking more than 80% of the study drug dosage prescribed for that interval. If a participant was nonadherent, both patient and caregiver were counseled on the importance of taking the prescribed amount of study medication. Participants and their caregivers were instructed to keep records of their exercise activity and heart rate. This information was also used to estimate adherence. Demographics, training, professional status of the investigators was not clearly mentioned. Metformin Class of drug: Behavioural + Pharmacological Dose: 250 mg, increased after 4 days to 750 mg. Frequency: Once a day, increased to 3 times a day Duration: 12 weeks Who delivered the intervention: Participant, monitored by parent/ caregiver Type of intervention: Pharmacological Lifestyle intervention + Placebo Class of drug: Behavioural intervention + Placebo Dose: ‐ Frequency: During 4 weeks + daily exercise Duration: 12 weeks Who delivered the intervention: Unclear Type of intervention: Behavioural + Placebo Placebo Class of drug: Placebo Dose: ‐ Frequency: Once a day, increased to 3 times a day Duration: 12 weeks Who delivered the intervention: Unclear Type of intervention: Placebo
Outcomes	Fasting blood sugar Outcome type: Continuous Reporting: Fully reported Unit of measure: mg/dL Direction: Lower is better Data value: Endpoint BMI Outcome type: Continuous Unit of measure: Kg/m2 Direction: Lower is better Data value: Endpoint Waist circumference Outcome type: Continuous Reporting: Fully reported Unit of measure: cm Direction: Lower is better Data value: Endpoint Drop‐out Outcome type: Adverse event Reporting: Fully reported Direction: Lower is better Data value: Endpoint
Identification	Sponsorship source: National Key Technologies R&D Program in the 10th 5‐year‐plan grant 2004BA720A22 from the Ministry of Science and Technology of the People’s Republic of China. Country: China Setting: Mental Health Institute Comments: ‐ Authors name: Ren‐Rong Wu Institution: Mental Health Institute of the Second Xiangya Hospital Email: wurenrong2005@yahoo.com.cn Address: Mental Health Institute of the Second Xiangya Hospital, Central South University, Mail Code: 410011,139 Renmin Middle Rd, Changsha, Hunan, China
Notes	Only metformin versus placebo data included in meta‐analyses. Results of other study arms described narratively.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "Participants were randomized through a computer‐generated table" Judgement comment: Random sequence generation by computer.
Allocation concealment (selection bias)	Low risk	Quote: "Assignment was determined after completing all screening assessments and being accepted into the study. To ensure concealment of the treatment assignment, randomization was conducted independently of the investigators by a research pharmacist" Judgement comment: allocation concealed from investigators.
Blinding of participants and personnel (performance bias) All outcomes	Low risk	"Metformin treatments were administered in a double‐blind placebocontrolled fashion." Judgement comment: Participants were blinded to medication/placebo.
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Judgement comment: As placebo was used, outcome assessors were blinded to placebo/medication but possibly not to behavioural intervention.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Judgement comment: No obvious difference in drop‐out between study arms. LOCF used for missing data; this may cause bias but unlikely due to low drop‐out rates.
Selective reporting (reporting bias)	Unclear risk	Judgement comment: Trial registered retrospectively.
Other sources of bias	Unclear risk	Judgement comment: Even though only participants who had gained more than 10% body weight on medication were included, patients were on average not overweight and mostly fitted within a very narrow BMI range of 24‐25 at baseline. We also had some queries about the data presented in Table 2, because several estimates and confidence intervals were identical for different groups and timepoints. The authors were contacted for more information twice in three weeks, but we received no response.